One or more multi-stage optimization iterations are performed with respect to a compression algorithm. A given iteration comprises a first stage in which hyper-parameters of a perceptual quality algorithm are tuned independently of the compression algorithm. A second stage of the iteration comprises tuning hyper-parameters of the compression algorithm using a set of perceptual quality scores generated by the tuned perceptual quality algorithm. The final stage of the iteration comprises performing a compression quality evaluation test on the tuned compression algorithm.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
2. The system as recited in claim 1, wherein input of the second fitness function with respect to the particular hyper-parameter combination comprises one or more of: (a) a perceptual quality score obtained for the image compressed using the particular hyper-parameter combination or (b) a size of the image compressed using the particular hyper-parameter combination.
The system optimizes image compression by evaluating hyper-parameter combinations using a second fitness function. This function assesses the quality and efficiency of compression for a given set of hyper-parameters. The evaluation includes at least one of two metrics: (a) a perceptual quality score, which measures how well the compressed image retains visual fidelity compared to the original, or (b) the file size of the compressed image, which indicates storage or transmission efficiency. The system uses these metrics to determine the optimal hyper-parameter settings for balancing quality and compression ratio. The first fitness function, referenced in the broader system, likely defines an initial set of criteria for evaluating hyper-parameters, such as computational efficiency or other performance metrics. The second fitness function refines this evaluation by incorporating perceptual quality and size, ensuring the final compression settings meet both technical and user-experience requirements. This approach allows for automated tuning of compression algorithms to achieve desired trade-offs between quality and file size.
4. The system as recited in claim 1, wherein the first image file format comprises one of: (a) a JPEG (Joint Photographic Experts Group) format, (b) a WebP format, (c) an AVIF format, or (d) an MP4 (Motion Pictures Experts Group-4) format.
The invention relates to a system for processing image files in various formats to optimize storage, transmission, or display. The system addresses the challenge of handling different image file formats efficiently, particularly in applications where compatibility, compression, or quality are critical. The system supports multiple image file formats, including JPEG (Joint Photographic Experts Group), WebP, AVIF, and MP4 (Motion Pictures Experts Group-4). JPEG is a widely used lossy compression format for photographs, balancing file size and quality. WebP is a modern format offering both lossy and lossless compression with smaller file sizes compared to JPEG. AVIF (AV1 Image File Format) leverages the AV1 video codec for superior compression efficiency, particularly for high-quality images. MP4 is a container format commonly used for video but can also store image sequences or single frames. The system processes these formats to ensure compatibility across different devices and platforms, optimize storage space, or enhance transmission speed without significant quality degradation. The inclusion of these formats allows the system to adapt to various use cases, such as web applications, multimedia storage, or real-time streaming, where different formats may be preferred based on performance requirements.
5. The system as recited in claim 1, wherein the one or more hyper-parameters of the first compression algorithm comprise a parameter associated with one or more of: (a) chroma subsampling, (b) block prediction, (c) frequency domain transformation, (d) quantization or (e) run-length encoding.
This invention relates to video compression systems, specifically improving compression efficiency by optimizing hyper-parameters in a first compression algorithm. The system addresses the challenge of balancing compression ratio and visual quality in video encoding by dynamically adjusting key hyper-parameters that influence compression performance. These parameters include chroma subsampling, which reduces color resolution to save bandwidth; block prediction, which improves spatial redundancy reduction; frequency domain transformation, which converts spatial data into frequency components for more efficient encoding; quantization, which controls bit allocation and distortion; and run-length encoding, which compresses repeated data patterns. By fine-tuning these parameters, the system enhances compression efficiency while maintaining acceptable visual quality. The invention is part of a broader system that may also include a second compression algorithm, where the first algorithm's output is further processed to achieve additional compression gains. The dynamic adjustment of these hyper-parameters allows the system to adapt to different video content types and encoding requirements, optimizing both computational efficiency and perceptual quality.
7. The method as recited in claim 6, wherein said tuning the one or more hyper-parameters of the first image compression algorithm comprises utilizing an evolutionary algorithm.
The invention relates to image compression techniques, specifically methods for optimizing hyper-parameters in image compression algorithms. The problem addressed is the need for efficient and adaptive compression methods that balance compression ratio and image quality. Traditional compression algorithms often rely on fixed hyper-parameters, which may not be optimal for varying image types or quality requirements. The method involves tuning hyper-parameters of a first image compression algorithm using an evolutionary algorithm. Evolutionary algorithms are optimization techniques inspired by natural selection, where candidate solutions evolve over generations to improve performance. In this context, the evolutionary algorithm iteratively adjusts the hyper-parameters of the compression algorithm to maximize compression efficiency while maintaining acceptable image quality. The process may include evaluating multiple candidate parameter sets, selecting the best-performing ones, and generating new candidates through operations like mutation and crossover. This adaptive approach allows the compression algorithm to dynamically adapt to different image characteristics, improving overall performance compared to static parameter settings. The method may also involve comparing the performance of the tuned algorithm against a second compression algorithm to select the most effective approach for a given application.
8. The method as recited in claim 7, wherein a fitness function used in the evolutionary algorithm is based at least in part on a penalty value, wherein the penalty value depends at least in part on a file size of a compressed file generated using the first image compression algorithm.
This invention relates to image compression techniques using evolutionary algorithms. The problem addressed is optimizing image compression algorithms to balance compression efficiency with file size constraints. The solution involves an evolutionary algorithm that evaluates and improves compression algorithms based on a fitness function incorporating a penalty value. This penalty value is derived from the file size of a compressed image, ensuring that the algorithm prioritizes smaller file sizes while maintaining acceptable compression quality. The evolutionary algorithm iteratively selects, mutates, and evaluates candidate compression algorithms, using the fitness function to guide the selection process. The penalty value dynamically adjusts based on the file size of the compressed output, allowing the algorithm to adapt to different compression requirements. This approach enables the discovery of optimized compression algorithms that meet specific file size targets while preserving image quality. The method is particularly useful in applications where storage or transmission efficiency is critical, such as digital media distribution or cloud storage systems. By integrating the penalty-based fitness function, the system ensures that the evolutionary algorithm efficiently explores the solution space to find the best-performing compression algorithms under given constraints.
9. The method as recited in claim 8, wherein the penalty value depends on a difference in perceptual quality scores between (a) the compressed file generated using the first image compression algorithm and (b) a compressed file generated using the reference compression algorithm.
The invention relates to image compression techniques, specifically improving compression efficiency while maintaining perceptual quality. The problem addressed is the trade-off between compression ratio and perceptual quality, where traditional methods may sacrifice visual fidelity for higher compression or vice versa. The invention introduces a method that optimizes compression by incorporating a penalty value based on perceptual quality differences. The method involves comparing two compressed versions of an image: one generated using a first image compression algorithm and another generated using a reference compression algorithm. The penalty value is determined by the difference in perceptual quality scores between these two compressed files. This penalty value is then used to adjust the compression process, ensuring that the first algorithm achieves a balance between compression efficiency and perceptual quality. The reference compression algorithm serves as a benchmark, providing a baseline for evaluating the perceptual impact of the first algorithm. By dynamically adjusting the compression parameters based on the penalty value, the method ensures that the compressed image retains acceptable perceptual quality while achieving optimal compression efficiency. This approach is particularly useful in applications where both file size and visual quality are critical, such as digital media storage, streaming, and transmission. The invention enhances existing compression techniques by introducing a feedback mechanism that prioritizes perceptual quality without compromising compression performance.
11. The method as recited in claim 6, wherein said tuning the one or more hyper-parameters of the first perceptual quality algorithm comprises utilizing an evolutionary algorithm.
This invention relates to optimizing perceptual quality algorithms in machine learning systems, particularly for tasks like image or audio processing. The core problem addressed is the challenge of efficiently tuning hyper-parameters in perceptual quality algorithms to improve performance without excessive computational cost. Traditional tuning methods often rely on manual adjustments or grid searches, which are time-consuming and may not yield optimal results. The invention describes a method that uses an evolutionary algorithm to automatically tune one or more hyper-parameters of a perceptual quality algorithm. Evolutionary algorithms are bio-inspired optimization techniques that mimic natural selection, iteratively improving solutions through processes like mutation, crossover, and selection. By applying this approach, the method efficiently explores the hyper-parameter space to find configurations that maximize the perceptual quality of the output, such as image sharpness or audio clarity. The method may also involve evaluating the tuned algorithm using a validation dataset to ensure robustness. This technique is particularly useful in applications where perceptual quality is critical, such as media compression, enhancement, or synthesis. The use of evolutionary algorithms allows for automated, data-driven optimization, reducing the need for manual intervention and improving the efficiency of the tuning process. The invention may be applied to various types of perceptual quality algorithms, including those used in deep learning models or traditional signal processing pipelines.
12. The method as recited in claim 11, wherein a fitness function used in the evolutionary algorithm used for tuning the one or more hyper-parameters of the first perceptual quality algorithm is based at least in part on a metric of disagreement between (a) intra-image-pair quality preferences indicated by one or more annotators and (b) corresponding intra-image-pair quality preferences generated by the perceptual quality algorithm.
This invention relates to improving perceptual quality algorithms through evolutionary optimization. The problem addressed is the challenge of accurately tuning hyper-parameters in perceptual quality algorithms, which are used to assess the visual quality of images or videos. Traditional tuning methods often rely on manual adjustments or simple optimization techniques, which may not efficiently capture the nuances of human perceptual preferences. The invention describes a method for tuning hyper-parameters of a perceptual quality algorithm using an evolutionary algorithm. The evolutionary algorithm iteratively optimizes the hyper-parameters based on a fitness function. The fitness function evaluates the algorithm's performance by comparing its output with human annotations. Specifically, it measures the disagreement between quality preferences indicated by human annotators for pairs of images and the corresponding preferences generated by the perceptual quality algorithm. By minimizing this disagreement, the method refines the algorithm to better align with human perceptual judgments. The evolutionary algorithm may include operations such as selection, crossover, and mutation to explore the hyper-parameter space. The method ensures that the tuned perceptual quality algorithm produces results that are more consistent with human evaluations, improving its reliability and accuracy in real-world applications. This approach is particularly useful in fields like image processing, computer vision, and multimedia quality assessment, where perceptual fidelity is critical.
15. The method as recited in claim 13, wherein said conducting the compression quality evaluation further comprises utilizing a mixed-effects model in which one or more of: (a) potential annotator bias or (b) image-specific offsets are modeled as respective random effects.
This invention relates to evaluating the quality of image compression, particularly in scenarios where human annotators assess compression artifacts. The problem addressed is the variability introduced by different annotators and image-specific factors, which can skew compression quality assessments. The solution involves using a mixed-effects model to account for these variations. The model treats annotator bias and image-specific offsets as random effects, allowing for more accurate and reliable quality evaluations. By incorporating these random effects, the method reduces the impact of individual annotator tendencies and inherent differences between images, leading to more consistent and objective compression quality measurements. This approach is particularly useful in applications where subjective human evaluation is combined with statistical analysis to assess compression performance. The mixed-effects model provides a robust framework for separating true compression artifacts from noise introduced by annotator variability or image characteristics, improving the reliability of quality metrics in image compression systems.
17. The one or more non-transitory computer-accessible storage media as recited in claim 16, wherein said tuning the one or more hyper-parameters of the first media object compression algorithm comprises utilizing an evolutionary algorithm.
This invention relates to optimizing media object compression algorithms using evolutionary algorithms. The problem addressed is the challenge of efficiently tuning hyper-parameters in media compression to balance quality and compression efficiency. Traditional methods often rely on manual tuning or heuristic approaches, which can be time-consuming and suboptimal. The invention involves a system that stores one or more media objects, such as images, videos, or audio files, and applies a first media object compression algorithm to these objects. The system further includes a second media object compression algorithm that serves as a reference for evaluating the performance of the first algorithm. The system compares the output of the first algorithm against the second algorithm to assess compression quality and efficiency. A key aspect of the invention is the use of an evolutionary algorithm to tune the hyper-parameters of the first compression algorithm. Evolutionary algorithms, inspired by natural selection, iteratively improve solutions by selecting, recombining, and mutating candidate parameter sets. This approach automates the optimization process, reducing the need for manual intervention and improving the likelihood of finding optimal or near-optimal hyper-parameter configurations. The system may also include a user interface for displaying compression results and allowing further adjustments. This method enhances compression performance while maintaining or improving media quality.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
May 15, 2020
December 13, 2022
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.